capture log close

log using "W:\Research\Current research\China\Analyses\Output\myfile.smcl", replace

*Initializing data
clear
do "W:\Research\Current research\China\Analyses\data_initializing.do"


******************************

*** TABLE OF CONTENTS

*** MAIN TABLES

*	Table 1: Descriptive statistics on firms
*	Panel A. Descriptive statistics on hard variables for firms
*	Panel B. Descriptive statistics on soft variables for firms and loan outcomes
*	Panel C. Distributions of overall recommendations, environmental scores, and safety risk scores

*	Table 2: Descriptive statistics on customer managers and loan officers
*	Panel A. Customer managers
*	Panel B. Loan officers

*	Table 3: Correlations
*	Panel A. Correlations between loan grant decision and firm soft information variables
*	Panel B. Correlations between customer managers' biospheric values and their evaluations
*	Panel C. Correlations between loan officers' biospheric values and firm‐level variables
*	Panel D. Correlations between customer managers' biospheric values and their other personal traits

*	Table 4: Modeling Overall Recommendations
*	Panel A. Coefficients and t-values
*	Panel B. Standardized coefficients

*	Table 5. Customer managers' biospheric-values–environmental-score interaction and their overall recommendations. 
*	Panel A. Coefficients and t-values by biospheric-values quintile
*	Panel B. Standardized coefficients by biospheric-values quintile
*	Panel C. Pooled regression interacting biospheric-values quintile dummies with environmental score. 

*	Table 6: Modeling Loan Granting Decisions

*	Table 7. Loan officers' biospheric-values–environmental-score interaction and their loan granting decision
*	Panel A. Coefficients and t-values by biospheric-values quintile
*	Panel B. Modeling the decisions of loan officers with biospheric-values interactions

*	Table 8. Loan officers' biospheric values and loan performance 
*	Panel A. Loan performance by loan officers' biospheric-values quintile
*	Panel B. Modeling loan performance with loan officers' biospheric values and interactions

*** APPENDIX TABLES

*	Table A1. Additional descriptive statistics on firms.
*	Panel A. Distribution of loan applicants by industry
*	Panel B. Distribution by company registration type
*	Panel C. Distribution of environmental scores by industry
*	Panel D. Distribution of truthfulness and reasonableness scores
*	Panel E. Distribution of repayment ability scores
*	Panel F. Distribution of repayment willingness scores

*	Table A2. Joint distribution of environmental score, safety score, and truthfulness score with overall recommendation and loan acceptance 
*	Panel A. Joint distribution of environmental score and overall recommendation
*	Panel B. Joint distribution of safety score and overall recommendation
*	Panel C. Joint distribution of truthfulness and reasonableness score and overall recommendation
*	Panel D. Joint distribution of environmental score and loan acceptance
*	Panel E. Joint distribution of safety score and loan acceptance
*	Panel F. Joint distribution of truthfulness and reasonableness score and loan acceptance

*	Table A3. Descriptive statistics on other bankers' biospheric values 

*	Table A4. Correlations between customer managers' biospheric values and the correlation of the environmental score with other scores

*	Table A5. Correlations between loan officers' biospheric values, gender, education, age, and experience 

*	Table A6. Modeling overall recommendations by emphasizing the role of the environmental score 

*	Table A7. Modeling the decisions of customer managers using biospheric-values interactions: ordered logit specification 

*	Table A8. Loan officers' biospheric-values–environmental-score interaction and their loan granting decision: standardized coefficients 

*	Table A9. Modeling the decisions of loan officers using biospheric-values interactions: logit specification 

*	Table A10. When do loan officers exercise discretion? 
*	Panel A. Number of rejected loan applications 
*	Panel B. Number of approved loan applications
*	Panel C. Loan officer discretion as a function of customer managers' overoptimism
*	Panel D. Correlation between customer managers' overoptimism and loan performance measures

*	Table A11. Loan officers' biospheric-values–environmental-score interaction and their loan granting decisions: subsample analysis by loan approval probability terciles 

*	Table A12. Loan outcomes for firms with male vs. female majority owners

*	Table A13. Gender bias vs. green bias among loan officers 

*	Table A14. Gender bias vs. green bias among customer managers 

*   Table A15. Additional descriptive statistics on customer managers and loan officers
*	Panel A. Customer managers
*	Panel B. Loan officers
*	Panel C. Correlations between customer managers' biospheric values and their other environmental traits
*	Panel D. Correlations between loan officers' biospheric values and their other environmental traits

*   Table A16. Distinguishing between taste-based and statistical discrimination in loan officers' decision-making 
*	Panel A. Biospheric-values interactions
*	Panel B. Green-beliefs interactions

*	Table A17. Loan granting decisions when loan officers hold below-median green beliefs 

*	Table A18. Placebo tests using alternative banker traits
*	Panel A. Customer managers' overall recommendations by gender
*	Panel B. Customer managers' overall recommendations by education
*	Panel C. Customer managers' overall recommendations by age
*	Panel D. Customer managers' overall recommendations by experience 

*	Table A19. Loan officers' biospheric values and loan performance: robustness

*	Table A20. Loan officers' biospheric values and loan performance: loans overdue by less than 90 days
*	Panel A. Loan performance by loan officers' biospheric-values quintile
*	Panel B. Modeling loan performance with loan officers' biospheric values and interactions

*	Table A21. Subcomponents of biospheric values
*	Panel A. Correlations between customer managers' mean biospheric values and its subcomponents
*	Panel B. Modeling customer managers' biospheric-values–environmental-score interaction using different measures

*	Table A22. Loan granting decisions by industry greenness terciles 
*	Panel A. Coefficients and t-values 
*	Panel B. Standardized coefficients

*	Table A23. Loan granting decisions by loan officer biospheric values quartiles 

*	Table A24. Loan outcomes conditional on loan approval 

*	Table A25. Treatment effects on environmental score, safety score, overall recommendation, and loan granting decision 

*   APPENDIX. Randomization tests

*** TABLE OF CONTENTS ENDS


********
* CODE
********

*** MAIN TABLES

*	Table 1: Descriptive statistics on firms

*	Panel A. Descriptive statistics on hard variables for firms
asdoc tabstat Firmage TotalAsset CurrentRatio_w DebtAssetRatio_w NetProfitRatio_w SalesGrowth_w, stat(mean median sd min max N) col(stat) replace label title(Descriptive statistics on hard variables on firms)

*	Panel B. Descriptive statistics on soft variables for firms and loan outcomes
asdoc tabstat TruthfulnessandReasonableness Repaymentability RepaymentWillingnes Envirorisk Safetyrisk Recommend Outcome Collateral ExecutiveAnnualInterestRate Maturity2 FinDefault, stat(mean median sd min max N) col(stat) append label title(Descriptive statistics on soft variables on firms and loan outcomes)
asdoc tabstat ApprovedAmount if Outcome==1, stat(mean median sd min max N) col(stat) append label title(Descriptive statistics on soft variables on firms and loan outcomes)

*	Panel C. Distributions of overall recommendations, environmental scores, and safety risk scores

* Distribution of overall recommendations
asdoc tab Recommend, append label title(Distribution of customer managers' recommendations) 

* Distribution of environmental scores
asdoc tab Envirorisk, append label title(Distribution of environmental scores) 

* Distribution of safety risk scores
asdoc tab Safetyrisk, append label title( Distribution of safety risk scores)  



*	Table 2: Descriptive statistics on customer managers and loan officers
*   Includes inputs also for Table A15 Panel A and B

*	Panel A. Customer managers

* Age, experience, biospheric values, standard deviation of environmental score, green values, green beliefs, and subjective green information of customer managers 
collapse (mean) C_Age C_ExperienceinBank C_Biosphericvalues C_Envrisk_variance C_Envvalues C_Envbeliefs C_Envinfo, by (C_ID)
asdoc tabstat C_Age C_ExperienceinBank C_Biosphericvalues C_Envrisk_variance C_Envvalues C_Envbeliefs C_Envinfo, stat(mean median sd skewness min max N) col(stat) label title(Distribution of customer managers' age, experience, biospheric values, standard deviation of environmental score, green values, green beliefs, and subjective green information) 

* Gender and education of customer managers 
clear all
do "W:\Research\Current research\China\Analyses\data_initializing.do"
collapse (firstnm) C_Gender C_Education, by (C_ID)
asdoc tab C_Gender, append label title(Distribution of customer managers' gender)
asdoc tab C_Education, append label title(Distribution of customer managers'  education)

*	Panel B. Loan officers

* Age, experience, biospheric values, green values, green beliefs, and subjective green information of loan officers
clear all
do "W:\Research\Current research\China\Analyses\data_initializing.do"
collapse (mean) L_Age L_ExperienceinBank L_Biosphericvalues L_Envvalues L_Envbeliefs L_Envinfo, by (L_ID)
asdoc tabstat L_Age L_ExperienceinBank L_Biosphericvalues L_Envvalues L_Envbeliefs L_Envinfo, stat(mean median sd skewness min max N) col(stat) label title(Distribution of loan officers' age, experience, biospheric values, green values, green beliefs, and subjective green information) 

* Gender and education of loan officers
clear all
do "W:\Research\Current research\China\Analyses\data_initializing.do"
collapse (firstnm) L_Gender L_Education, by (L_ID)
asdoc tab L_Gender, append label title(Distribution of loan officers' gender)
asdoc tab L_Education, append label title(Distribution oloan officers'  education)



*	Table 3: Correlations

*	Panel A. Correlations between loan grant decision and firm soft information variables (N = 1436)
clear all
do "W:\Research\Current research\China\Analyses\data_initializing.do"

asdoc pwcorr Outcome Recommend $oldsofts Envirorisk Safetyrisk, sig append label title(Correlations between outcomes and soft firm variables)

*	Panel B. Correlations between customer managers' biospheric values and their evaluations (N = 202)
collapse (mean) Recommend Envirorisk C_Envvalues C_Envbeliefs C_Envinfo C_Biosphericvalues C_Envrisk_variance C_Cor_env_recommend, by (C_ID)
asdoc pwcorr C_Biosphericvalues Recommend Envirorisk C_Envrisk_variance C_Cor_env_recommend, sig append label title(Correlations between customer managers' biosphreric values and their overall recommendation decisions)

*	Panel C. Correlations between loan officers' biospheric values and firm‐level variables (N = 64)
clear all
do "W:\Research\Current research\China\Analyses\data_initializing.do"
collapse (mean) L_Biosphericvalues Outcome Envirorisk L_Cor_outcome_env, by (L_ID)
asdoc pwcorr L_Biosphericvalues Outcome Envirorisk L_Cor_outcome_env, sig append label title(Correlations between loan officers' biosphreric values and their loan grant decisions)

*	Panel D. Correlations between customer managers' biospheric values and their other personal traits
clear all
do "W:\Research\Current research\China\Analyses\data_initializing.do"
collapse (firstnm) C_Biosphericvalues C_Age C_Gender C_Edulevel C_ExperienceinBank, by (C_ID)
gen C_Female = 1
replace C_Female = 0 if C_Gender=="Male"
asdoc pwcorr C_Biosphericvalues C_Age C_Female C_Edulevel C_ExperienceinBank, sig 



*	Table 4: Modeling Overall Recommendations

*	Panel A. Coefficients and t-values

* Only hard variables
clear all
do "W:\Research\Current research\China\Analyses\data_initializing.do"

regress Recommend $lpm_controls, vce(cluster C_ID) 
outreg2 using T4a.xls, stats (coef tstat) bdec(3) tdec(2) addtext(Industry FE, Yes, Firm type FE, Yes) drop(i.Companyregtype i.Industrycode) nocons

* Environmental risk discarded
regress Recommend $lpm_controls $oldsofts Safetyrisk , vce(cluster C_ID) 
outreg2 using T4a.xls, stats (coef tstat) bdec(3) tdec(2) addtext(Industry FE, Yes, Firm type FE, Yes) drop(i.Companyregtype i.Industrycode) nocons

* Environmental risk score treated as continuous variable
regress Recommend $lpm_controls $oldsofts Safetyrisk Envirorisk, vce(cluster C_ID) 
outreg2 using T4a.xls, stats (coef tstat) bdec(3) tdec(2) addtext(Industry FE, Yes, Firm type FE, Yes) drop(i.Companyregtype i.Industrycode) nocons

* Environmental risk score split into dummies
regress Recommend $lpm_controls $oldsofts Safetyrisk i.Envirorisk, vce(cluster C_ID) 
outreg2 using T4a.xls, stats (coef tstat) bdec(3) tdec(2)  addtext(Industry FE, Yes, Firm type FE, Yes) drop(i.Companyregtype i.Industrycode) nocons

* Environmental risk score split into dummies, adding customer manager FEs
regress Recommend $lpm_controls $oldsofts Safetyrisk i.Envirorisk i.C_ID, vce(cluster C_ID) 
outreg2 using T4a.xls, stats (coef tstat) bdec(3) tdec(2) addtext(Industry FE, Yes, Firm type FE, Yes, Customer manager FE, Yes) drop(i.Companyregtype i.Industrycode i.C_ID) nocons

*	Panel B. Standardized coefficients

* Only hard variables
regress Recommend $lpm_controls, vce(r) beta
outreg2 using T4b.xls, stats (beta) bdec(3)  addtext(Industry FE, Yes, Firm type FE, Yes) drop(i.Companyregtype i.Industrycode) nocons

* Environmental risk discarded
regress Recommend $lpm_controls $oldsofts Safetyrisk , vce(r) beta
outreg2 using T4b.xls, stats (beta) bdec(3)  addtext(Industry FE, Yes, Firm type FE, Yes) drop(i.Companyregtype i.Industrycode) nocons

* Environmental risk score treated as continuous variable
regress Recommend $lpm_controls $oldsofts Safetyrisk Envirorisk, vce(r) beta
outreg2 using T4b.xls, stats (beta) bdec(3)  addtext(Industry FE, Yes, Firm type FE, Yes) drop(i.Companyregtype i.Industrycode) nocons

* Environmental risk score split into dummies
regress Recommend $lpm_controls $oldsofts Safetyrisk i.Envirorisk, vce(r) beta
outreg2 using T4b.xls, stats (beta) bdec(3)   addtext(Industry FE, Yes, Firm type FE, Yes) drop(i.Companyregtype i.Industrycode) nocons

* Environmental risk score split into dummies, adding customer manager FEs
regress Recommend $lpm_controls $oldsofts Safetyrisk i.Envirorisk i.C_ID, vce(r) beta
outreg2 using T4b.xls, stats (beta) bdec(3)  addtext(Industry FE, Yes, Firm type FE, Yes, Customer manager FE, Yes) drop(i.Companyregtype i.Industrycode i.C_ID) nocons



*	Table 5. Customer managers' biospheric-values–environmental-score interaction and their overall recommendations. 

*	Panel A. Coefficients and t-values by biospheric-values quintile

* For creation of table
reg Recommend $lpm_controls $oldsofts Safetyrisk Envirorisk if C_BioQ==0 , vce(cluster C_ID) 
outreg2 using T5a.xls, stats (coef tstat) bdec(3) tdec(2) addtext(Industry FE, Yes, Firm type FE, Yes)  drop(i.Companyregtype i.Industrycode) nocons
reg Recommend $lpm_controls $oldsofts Safetyrisk Envirorisk if C_BioQ==1 , vce(cluster C_ID) 
outreg2 using T5a.xls, stats (coef tstat) bdec(3) tdec(2) addtext(Industry FE, Yes, Firm type FE, Yes)  drop(i.Companyregtype i.Industrycode) nocons
reg Recommend $lpm_controls $oldsofts Safetyrisk Envirorisk if C_BioQ==2 , vce(cluster C_ID) 
outreg2 using T5a.xls, stats (coef tstat) bdec(3) tdec(2) addtext(Industry FE, Yes, Firm type FE, Yes)  drop(i.Companyregtype i.Industrycode) nocons
reg Recommend $lpm_controls $oldsofts Safetyrisk Envirorisk if C_BioQ==3 , vce(cluster C_ID) 
outreg2 using T5a.xls, stats (coef tstat) bdec(3) tdec(2) addtext(Industry FE, Yes, Firm type FE, Yes)  drop(i.Companyregtype i.Industrycode) nocons
reg Recommend $lpm_controls $oldsofts Safetyrisk Envirorisk if C_BioQ==4 , vce(cluster C_ID) 
outreg2 using T5a.xls, stats (coef tstat) bdec(3) tdec(2) addtext(Industry FE, Yes, Firm type FE, Yes)  drop(i.Companyregtype i.Industrycode) nocons 				

* Analysis continues: p-values for difference in key variables between high and low biospheric value quintiles

* Regressions for tests of differences in coefficients; drop clustering at the first stage
* The following fvset command is needed 
fvset base 1 Companyregtype

* Lowest customer manager biospheric values quintile
reg Recommend $lpm_controls $oldsofts Safetyrisk Envirorisk if C_BioQ==0 
estimates store C_BioQ0

* Lowest customer manager biospheric values quintile
reg Recommend $lpm_controls $oldsofts Safetyrisk Envirorisk if C_BioQ==4  
estimates store C_BioQ4
				
* Perform tests of differences for different variables
suest C_BioQ0 C_BioQ4, vce(cluster C_ID)
asdoc test [C_BioQ0_mean]Envirorisk = [C_BioQ4_mean]Envirorisk, append label title(Test of equality of customer managers' environmental risk coefficients, difference between top and bottom bispheric value quintiles)
capture log using "W:\Research\Current research\China\Analyses\Output\myfile.smcl", text append
asdoc test [C_BioQ0_mean]Safetyrisk = [C_BioQ4_mean]Safetyrisk, append label title(Test of equality of customer managers' safety risk coefficients, difference between top and bottom bispheric value quintiles)
capture log using "W:\Research\Current research\China\Analyses\Output\myfile.smcl", text append
asdoc test [C_BioQ0_mean]RepaymentWillingness = [C_BioQ4_mean]RepaymentWillingness, append label title(Test of equality of customer managers' repayment willingness coefficients, difference between top and bottom bispheric value quintiles)
capture log using "W:\Research\Current research\China\Analyses\Output\myfile.smcl", text append	
asdoc test [C_BioQ0_mean]TruthfulnessandReasonableness = [C_BioQ4_mean]TruthfulnessandReasonableness, append label title(Test of equality of customer managers' truthfullness coefficients, difference between top and bottom bispheric value quintiles)	
capture log using "W:\Research\Current research\China\Analyses\Output\myfile.smcl", text append
asdoc test [C_BioQ0_mean]Repaymentability = [C_BioQ4_mean]Repaymentability, append label title(Test of equality of customer managers' repaymentability coefficients, difference between top and bottom bispheric value quintiles)
capture log using "W:\Research\Current research\China\Analyses\Output\myfile.smcl", text append
asdoc test [C_BioQ0_mean]Firmage = [C_BioQ4_mean]Firmage, append label title(Test of equality of customer managers' firm age coefficients, difference between top and bottom bispheric value quintiles)
capture log using "W:\Research\Current research\China\Analyses\Output\myfile.smcl", text append
asdoc test [C_BioQ0_mean]DebtAssetRatio_w = [C_BioQ4_mean]DebtAssetRatio_w, append label title(Test of equality of customer managers' debt-to-asset ratio coefficients, difference between top and bottom bispheric value quintiles)
capture log using "W:\Research\Current research\China\Analyses\Output\myfile.smcl", text append
asdoc test [C_BioQ0_mean]logTotalAsset = [C_BioQ4_mean]logTotalAsset, append label title(Test of equality of customer managers' log total asset coefficients, difference between top and bottom bispheric value quintiles)
capture log using "W:\Research\Current research\China\Analyses\Output\myfile.smcl", text append
asdoc test [C_BioQ0_mean]CurrentRatio_w = [C_BioQ4_mean]CurrentRatio_w, append label title(Test of equality of customer managers' crrent ratio coefficients, difference between top and bottom bispheric value quintiles)
capture log using "W:\Research\Current research\China\Analyses\Output\myfile.smcl", text append
asdoc test [C_BioQ0_mean]NetProfitRatio_w = [C_BioQ4_mean]NetProfitRatio_w, append label title(Test of equality of customer managers' net profit ratios coefficients, difference between top and bottom bispheric value quintiles)
capture log using "W:\Research\Current research\China\Analyses\Output\myfile.smcl", text append
asdoc test [C_BioQ0_mean]SalesGrowth_w = [C_BioQ4_mean]SalesGrowth_w, append label title(Test of equality of customer managers' sales growth coefficients, difference between top and bottom bispheric value quintiles)
capture log using "W:\Research\Current research\China\Analyses\Output\myfile.smcl", text append
		
*	Panel B. Standardized coefficients by biospheric-values quintile

* No robust clustering for standardized coefficients 
reg Recommend $lpm_controls $oldsofts Safetyrisk Envirorisk if C_BioQ==0 , vce(r) beta
outreg2 using T5b.xls, stats (beta ) bdec(3) addtext(Industry FE, Yes, Firm type FE, Yes) drop(i.Companyregtype i.Industrycode) nocons
reg Recommend $lpm_controls $oldsofts Safetyrisk Envirorisk if C_BioQ==1 , vce(r) beta
outreg2 using T5b.xls, stats (beta ) bdec(3) addtext(Industry FE, Yes, Firm type FE, Yes) drop(i.Companyregtype i.Industrycode) nocons
reg Recommend $lpm_controls $oldsofts Safetyrisk Envirorisk if C_BioQ==2 , vce(r) beta
outreg2 using T5b.xls, stats (beta ) bdec(3) addtext(Industry FE, Yes, Firm type FE, Yes) drop(i.Companyregtype i.Industrycode) nocons
reg Recommend $lpm_controls $oldsofts Safetyrisk Envirorisk if C_BioQ==3 , vce(r) beta
outreg2 using T5b.xls, stats (beta ) bdec(3) addtext(Industry FE, Yes, Firm type FE, Yes) drop(i.Companyregtype i.Industrycode) nocons
reg Recommend $lpm_controls $oldsofts Safetyrisk Envirorisk if C_BioQ==4 , vce(r) beta
outreg2 using T5b.xls, stats (beta ) bdec(3) addtext(Industry FE, Yes, Firm type FE, Yes) drop(i.Companyregtype i.Industrycode) nocons		

*	Panel C. Pooled regression interacting biospheric-values quintile dummies with environmental score

* Continuous environmental risk, centering
qui fvset base 4 C_BioQ
regress Recommend $lpm_controls $oldsofts Safetyrisk i.C_BioQ C.Envirorisk_center#i.C_BioQ, vce(cluster C_ID) 
outreg2 using T5c.xls, stats (coef tstat) bdec(3) tdec(2) addtext(Firm controls, Yes, Other Softs, Yes)  drop($lpm_controls  $oldsofts Safetyrisk) nocons
regress Recommend $lpm_controls $oldsofts Safetyrisk C.Envirorisk_center#i.C_BioQ i.C_ID, vce(cluster C_ID) 
outreg2 using T5c.xls, stats (coef tstat) bdec(3) tdec(2) addtext(Firm controls, Yes, Other Softs, Yes, Customer manager FE, Yes)  drop($lpm_controls  $oldsofts Safetyrisk i.C_ID) nocons



*	Table 6: Modeling Loan Granting Decisions

* Environmental risk discarded
regress Outcome $lpm_controls Recommend $oldsofts Safetyrisk , vce(cluster L_ID) 
outreg2 using T6.xls, stats (coef tstat) bdec(3) tdec(2) addtext(Industry FE, Yes, Firm type FE, Yes) drop(i.Companyregtype i.Industrycode) nocons

* Environmental risk score treated as continuous variable
regress Outcome $lpm_controls Recommend $oldsofts Safetyrisk Envirorisk, vce(cluster L_ID) 
outreg2 using T6.xls, stats (coef tstat) bdec(3) tdec(2) addtext(Industry FE, Yes, Firm type FE, Yes) drop(i.Companyregtype i.Industrycode) nocons

* Environmental risk score split into dummies
regress Outcome $lpm_controls Recommend $oldsofts Safetyrisk i.Envirorisk, vce(cluster L_ID) 
outreg2 using T6.xls, stats (coef tstat) bdec(3) tdec(2) addtext(Industry FE, Yes, Firm type FE, Yes) drop(i.Companyregtype i.Industrycode) nocons

* Environmental risk score split into dummies, adding loan officer FEs
regress Outcome $lpm_controls Recommend $oldsofts Safetyrisk i.Envirorisk  i.L_ID, vce(cluster L_ID) 
outreg2 using T6.xls, stats (coef tstat) bdec(3) tdec(2) addtext(Industry FE, Yes, Firm type FE, Yes, Loan officer FE, Yes) drop(i.Companyregtype i.Industrycode i.L_ID) nocons



*	Table 7. Loan officers' biospheric-values–environmental-score interaction and their loan granting decision

*	Panel A. Coefficients and t-values by biospheric-values quintile

reg Outcome $lpm_controls Recommend $oldsofts Safetyrisk Envirorisk if L_BioQ==0 , vce(cluster L_ID) 
outreg2 using T7a.xls, stats (coef tstat) bdec(3) tdec(2) addtext(Industry FE, Yes, Firm type FE, Yes)  drop(i.Companyregtype i.Industrycode) nocons
reg Outcome $lpm_controls Recommend $oldsofts Safetyrisk Envirorisk if L_BioQ==1 , vce(cluster L_ID) 
outreg2 using T7a.xls, stats (coef tstat) bdec(3) tdec(2) addtext(Industry FE, Yes, Firm type FE, Yes)  drop(i.Companyregtype i.Industrycode) nocons
reg Outcome $lpm_controls Recommend $oldsofts Safetyrisk Envirorisk if L_BioQ==2 , vce(cluster L_ID) 
outreg2 using T7a.xls, stats (coef tstat) bdec(3) tdec(2) addtext(Industry FE, Yes, Firm type FE, Yes)  drop(i.Companyregtype i.Industrycode) nocons
reg Outcome $lpm_controls Recommend $oldsofts Safetyrisk Envirorisk if L_BioQ==3 , vce(cluster L_ID) 
outreg2 using T7a.xls, stats (coef tstat) bdec(3) tdec(2) addtext(Industry FE, Yes, Firm type FE, Yes)  drop(i.Companyregtype i.Industrycode) nocons
reg Outcome $lpm_controls Recommend $oldsofts Safetyrisk Envirorisk if L_BioQ==4 , vce(cluster L_ID) 
outreg2 using T7a.xls, stats (coef tstat) bdec(3) tdec(2) addtext(Industry FE, Yes, Firm type FE, Yes)  drop(i.Companyregtype i.Industrycode) nocons

* Analysis continues: p-values for difference in key variables between high and low biospheric value quintiles

* Regressions for tests of differences in coefficients; drop clustering at the first stage
* The following fvset command is needed 
fvset base 1 Companyregtype

* Lowest loan officer biospheric values quintile
reg  Outcome $lpm_controls Recommend $oldsofts Safetyrisk Envirorisk if L_BioQ==0 
estimates store L_BioQ0

* Highest loan officer biospheric values quintile
reg  Outcome $lpm_controls Recommend $oldsofts Safetyrisk Envirorisk if L_BioQ==4 
estimates store L_BioQ4

* Clustering comes into play here				
suest L_BioQ0 L_BioQ4, vce(cluster L_ID)

* Perform tests of differences for different variables
asdoc test [L_BioQ0_mean]Envirorisk = [L_BioQ4_mean]Envirorisk, append label title(Test of equality of loan officers' environmental risk coefficients, difference between top and bottom bispheric value quintiles)
capture log using "W:\Research\Current research\China\Analyses\Output\myfile.smcl", text append
asdoc test [L_BioQ0_mean]Safetyrisk = [L_BioQ4_mean]Safetyrisk, append label title(Test of equality of loan officers' safety risk coefficients, difference between top and bottom bispheric value quintiles)
capture log using "W:\Research\Current research\China\Analyses\Output\myfile.smcl", text append
asdoc test [L_BioQ0_mean]Recommend = [L_BioQ4_mean]Recommend, append label title(Test of equality of loan officers' recommend coefficients, difference between top and bottom bispheric value quintiles)
capture log using "W:\Research\Current research\China\Analyses\Output\myfile.smcl", text append
asdoc test [L_BioQ0_mean]RepaymentWillingness = [L_BioQ4_mean]RepaymentWillingness, append label title(Test of equality of loan officers' repayment willingness coefficients, difference between top and bottom bispheric value quintiles)
capture log using "W:\Research\Current research\China\Analyses\Output\myfile.smcl", text append	
asdoc test [L_BioQ0_mean]TruthfulnessandReasonableness = [L_BioQ4_mean]TruthfulnessandReasonableness, append label title(Test of equality of loan officers' truthfullness coefficients, difference between top and bottom bispheric value quintiles)
capture log using "W:\Research\Current research\China\Analyses\Output\myfile.smcl", text append	
asdoc test [L_BioQ0_mean]Repaymentability = [L_BioQ4_mean]Repaymentability, append label title(Test of equality of loan officers' repaymentability coefficients, difference between top and bottom bispheric value quintiles)
capture log using "W:\Research\Current research\China\Analyses\Output\myfile.smcl", text append
asdoc test [L_BioQ0_mean]Firmage = [L_BioQ4_mean]Firmage, append label title(Test of equality of loan officers' firm age coefficients, difference between top and bottom bispheric value quintiles)
capture log using "W:\Research\Current research\China\Analyses\Output\myfile.smcl", text append
asdoc test [L_BioQ0_mean]DebtAssetRatio_w = [L_BioQ4_mean]DebtAssetRatio_w, append label title(Test of equality of loan officers' debt-to-asset ratio coefficients, difference between top and bottom bispheric value quintiles)
capture log using "W:\Research\Current research\China\Analyses\Output\myfile.smcl", text append
asdoc test [L_BioQ0_mean]logTotalAsset = [L_BioQ4_mean]logTotalAsset, append label title(Test of equality of loan officers' log total asset coefficients, difference between top and bottom bispheric value quintiles)
capture log using "W:\Research\Current research\China\Analyses\Output\myfile.smcl", text append
asdoc test [L_BioQ0_mean]CurrentRatio_w = [L_BioQ4_mean]CurrentRatio_w, append label title(Test of equality of loan officers' crrent ratio coefficients, difference between top and bottom bispheric value quintiles)
capture log using "W:\Research\Current research\China\Analyses\Output\myfile.smcl", text append
asdoc test [L_BioQ0_mean]NetProfitRatio_w = [L_BioQ4_mean]NetProfitRatio_w, append label title(Test of equality of loan officers' net profit ratios coefficients, difference between top and bottom bispheric value quintiles)
capture log using "W:\Research\Current research\China\Analyses\Output\myfile.smcl", text append
asdoc test [L_BioQ0_mean]SalesGrowth_w = [L_BioQ4_mean]SalesGrowth_w, append label title(Test of equality of loan officers' sales growth coefficients, difference between top and bottom bispheric value quintiles)
capture log using "W:\Research\Current research\China\Analyses\Output\myfile.smcl", text append				



*************************
* Temporary analysis starts
* Table A8: Standardized coefficients by biospheric-values quintile

* No robust clustering for standardized coeeficients 
reg Outcome $lpm_controls Recommend $oldsofts Safetyrisk Envirorisk if L_BioQ==0 , vce(r) beta
outreg2 using A8.xls, stats (beta ) bdec(3) addtext(Industry FE, Yes, Firm type FE, Yes) drop(i.Companyregtype i.Industrycode) nocons
reg Outcome $lpm_controls Recommend $oldsofts Safetyrisk Envirorisk if L_BioQ==1 , vce(r) beta
outreg2 using A8.xls, stats (beta ) bdec(3) addtext(Industry FE, Yes, Firm type FE, Yes) drop(i.Companyregtype i.Industrycode) nocons
reg Outcome $lpm_controls Recommend $oldsofts Safetyrisk Envirorisk if L_BioQ==2 , vce(r) beta
outreg2 using A8.xls, stats (beta ) bdec(3) addtext(Industry FE, Yes, Firm type FE, Yes) drop(i.Companyregtype i.Industrycode) nocons
reg Outcome $lpm_controls Recommend $oldsofts Safetyrisk Envirorisk if L_BioQ==3 , vce(r) beta
outreg2 using A8.xls, stats (beta ) bdec(3) addtext(Industry FE, Yes, Firm type FE, Yes) drop(i.Companyregtype i.Industrycode) nocons
reg Outcome $lpm_controls Recommend $oldsofts Safetyrisk Envirorisk if L_BioQ==4 , vce(r) beta
outreg2 using A8.xls, stats (beta ) bdec(3) addtext(Industry FE, Yes, Firm type FE, Yes) drop(i.Companyregtype i.Industrycode) nocons			

* Temporary analysis ends
***************************		



*	Panel B. Modeling the decisions of loan officers with biospheric-values interactions

* Continous environmental risk, centering
qui fvset base 4 L_BioQ
regress Outcome $lpm_controls Recommend $oldsofts Safetyrisk i.L_BioQ C.Envirorisk_center#i.L_BioQ, vce(cluster L_ID) 
outreg2 using T7b.xls, stats (coef tstat) bdec(3) tdec(2) addtext(Firm controls, Yes, Other Softs, Yes)  drop($lpm_controls Recommend $oldsofts Safetyrisk) nocons
regress Outcome $lpm_controls Recommend $oldsofts Safetyrisk C.Envirorisk_center#i.L_BioQ i.L_ID, vce(cluster L_ID) 
outreg2 using T7b.xls, stats (coef tstat) bdec(3) tdec(2) addtext(Firm controls, Yes, Other Softs, Yes, Loan officer FE, Yes)  drop($lpm_controls Recommend $oldsofts Safetyrisk i.L_ID) nocons




*	Table 8. Loan officers' biospheric values and loan performance 

*	Panel A. Loan performance by loan officers' biospheric-values quintile
asdoc tab L_BioQ FinDefault, append label title(Joint distribution of bad loans by loan officers' biospheric values) 

*	Panel B. Modeling loan performance with loan officers' biospheric values and interactions
clear
do "W:\Research\Current research\China\Analyses\data_initializing.do"
qui fvset base 0 L_BioQ

* Basic regression, environmental score treated as continuous variable
regress FinDefault $lpm_controls Recommend $oldsofts Safetyrisk Envirorisk Collateral Maturity2 ExecutiveAnnualInterestRate, vce(cluster L_ID) 
outreg2 using T8b.xls, stats (coef tstat) bdec(3) tdec(2) addtext(Industry FE, Yes, Firm type FE, Yes) drop(i.Companyregtype i.Industrycode) nocons

* Add loan officer biospheric value quintiles, environmental score treated as continuous variable
regress FinDefault $lpm_controls Recommend $oldsofts Safetyrisk Envirorisk Collateral Maturity2 ExecutiveAnnualInterestRate i.L_BioQ, vce(cluster L_ID) 
outreg2 using T8b.xls, stats (coef tstat) bdec(3) tdec(2) addtext(Industry FE, Yes, Firm type FE, Yes) drop(i.Companyregtype i.Industrycode) nocons

* Environmental risk score interacted with loan officer biospheric value quintile
regress FinDefault $lpm_controls Recommend $oldsofts Safetyrisk Collateral Maturity2 ExecutiveAnnualInterestRate i.L_BioQ C.Envirorisk_center#i.L_BioQ, vce(cluster L_ID) 
outreg2 using T8b.xls, stats (coef tstat) bdec(3) tdec(2) addtext(Industry FE, Yes, Firm type FE, Yes) drop(i.Companyregtype i.Industrycode) nocons

* Environmental risk score interacted with loan officer biospheric value quintile, adding loan officer FEs
regress FinDefault $lpm_controls Recommend $oldsofts Safetyrisk Collateral Maturity2 ExecutiveAnnualInterestRate C.Envirorisk_center#i.L_BioQ i.L_ID, vce(cluster L_ID) 
outreg2 using T8b.xls, stats (coef tstat) bdec(3) tdec(2) addtext(Industry FE, Yes, Firm type FE, Yes, Loan officer FE, Yes) drop(i.Companyregtype i.Industrycode i.L_ID) nocons





*** APPENDIX TABLES

*	Table A1. Additional descriptive statistics on firms

*	Panel A. Distribution of loan applicants by industry
asdoc tab Industrycode, append label title(Distribution of sample by industry)

*	Panel B. Distribution by company registration type
asdoc tab Companyregtype, append label title(Distribution of sample by company registration type) 

*	Panel C. Distribution of environmental scores by industry
tab2 Envirorisk Industrycode

*	Panel D. Distribution of truthfulness and reasonableness scores
asdoc tab TruthfulnessandReasonableness, append label title( Distribution of truthfulness and reasonableness scores)  

*	Panel E. Distribution of repayment ability scores
asdoc tab Repaymentability, append label title( Distribution of repayment ability scores)  

*	Panel F. Distribution of repayment willingness scores
asdoc tab RepaymentWillingness, append label title( Distribution of repayment willingness scores)  




*	Table A2. Joint distribution of environmental score, safety score, and truthfulness score with overall recommendation and loan acceptance 

*	Panel A. Joint distribution of environmental score and overall recommendation
asdoc tab2 Envirorisk Recommend

*	Panel B. Joint distribution of safety score and overall recommendation
asdoc tab2 Safetyrisk Recommend

*	Panel C. Joint distribution of truthfulness and reasonableness score and overall recommendation
asdoc tab2 TruthfulnessandReasonableness Recommend

*	Panel D. Joint distribution of environmental score and loan acceptance
asdoc tab2 Envirorisk Outcome

*	Panel E. Joint distribution of safety score and loan acceptance
asdoc tab2 Safetyrisk Outcome

*	Panel F. Joint distribution of truthfulness and reasonableness score and loan acceptance
asdoc tab2 TruthfulnessandReasonableness Outcome



*	Table A3. Descriptive statistics on other bankers' biospheric values 
clear all
import excel "W:\Research\Current research\China\Data from the bank\Survey results cleaned.xlsx", sheet("Data") firstrow

gen Greenval = (Respectingtheearth + Unitywithnature + Protectingtheenvironment + Preventingpollution) / 4

asdoc tabstat Greenval if (Customermanager + Loanofficer)==0, stat(mean median sd skewness min max N) col(stat)
asdoc tabstat Greenval if Position=="Accounting Department", stat(mean median sd skewness min max N) col(stat)  
asdoc tabstat Greenval if Position=="Audit Department", stat(mean median sd skewness min max N) col(stat)  
asdoc tabstat Greenval if Position=="Bank Asset Preservation Department", stat(mean median sd skewness min max N) col(stat)  
asdoc tabstat Greenval if Position=="Bank Product Management Department", stat(mean median sd skewness min max N) col(stat)  
asdoc tabstat Greenval if Position=="Bank Teller", stat(mean median sd skewness min max N) col(stat)  
asdoc tabstat Greenval if Position=="Credit Card Centre", stat(mean median sd skewness min max N) col(stat)  
asdoc tabstat Greenval if Position=="General Office", stat(mean median sd skewness min max N) col(stat)  
asdoc tabstat Greenval if Position=="Human Resources Department", stat(mean median sd skewness min max N) col(stat)  
asdoc tabstat Greenval if Position=="Interbank Department", stat(mean median sd skewness min max N) col(stat)  
asdoc tabstat Greenval if Position=="Legal Compliance Department", stat(mean median sd skewness min max N) col(stat)  
asdoc tabstat Greenval if Position=="Private Finance Department", stat(mean median sd skewness min max N) col(stat)  
asdoc tabstat Greenval if Position=="Purchasing Department", stat(mean median sd skewness min max N) col(stat)  
asdoc tabstat Greenval if Position=="Technology Operations Department", stat(mean median sd skewness min max N) col(stat)  



*	Table A4. Correlations between customer managers' biospheric values and the correlation of the environmental score with other scores
clear all
do "W:\Research\Current research\China\Analyses\data_initializing.do"
collapse (mean) C_Biosphericvalues C_Cor_env_recommend C_Cor_env_truthful C_Cor_env_repaymentability C_Cor_env_willingness C_Cor_env_safety, by (C_ID)
asdoc pwcorr C_Biosphericvalues C_Cor_env_recommend C_Cor_env_truthful C_Cor_env_repaymentability C_Cor_env_willingness C_Cor_env_safety, sig append label title(Correlations between customer managers' biospheric values and the correlation of the environmental score with other scores)




*	Table A5. Correlations between loan officers' biospheric values, gender, education, age, and experience 
clear all
do "W:\Research\Current research\China\Analyses\data_initializing.do"
collapse (firstnm) L_Biosphericvalues L_Age L_Gender L_Edulevel L_ExperienceinBank, by (L_ID)
gen L_Female = 1
replace L_Female = 0 if L_Gender=="Male"
asdoc pwcorr L_Biosphericvalues L_Age L_Female L_Edulevel L_ExperienceinBank, sig 



*	Table A6. Modeling overall recommendations by emphasizing the role of the environmental score 

* Environmental score only, continuous specification
clear all
do "W:\Research\Current research\China\Analyses\data_initializing.do"

regress Recommend Envirorisk, vce(cluster C_ID) 
outreg2 using A6.xls, stats (coef tstat) bdec(3) tdec(2) addtext(Industry FE, No, Firm type FE, No) nocons

* Add industry FE
regress Recommend Envirorisk i.Industrycode, vce(cluster C_ID) 
outreg2 using A6.xls, stats (coef tstat) bdec(3) tdec(2) addtext(Industry FE, Yes, Firm type FE, No) drop(i.Industrycode) nocons

* Add other hard variables
regress Recommend $lpm_controls Envirorisk, vce(cluster C_ID) 
outreg2 using A6.xls, stats (coef tstat) bdec(3) tdec(2) addtext(Industry FE, Yes, Firm type FE, Yes) drop(i.Companyregtype i.Industrycode) nocons

* Add soft variables
regress Recommend $lpm_controls $oldsofts Safetyrisk Envirorisk, vce(cluster C_ID) 
outreg2 using A6.xls, stats (coef tstat) bdec(3) tdec(2) addtext(Industry FE, Yes, Firm type FE, Yes) drop(i.Companyregtype i.Industrycode) nocons



*	Table A7. Modeling the decisions of customer managers using biospheric-values interactions: ordered logit specification 
 
* Customer manager environmental value quintiles, continous environmental risk, centering

qui fvset base 4 C_BioQ
ologit Recommend $lpm_controls $oldsofts Safetyrisk i.C_BioQ C.Envirorisk_center#i.C_BioQ, vce(cluster C_ID) 
outreg2 using A7.xls, stats (coef tstat) bdec(3) tdec(2) addtext(Firm controls, Yes, Other Softs, Yes)  drop($lpm_controls  $oldsofts Safetyrisk) nocons
 
ologit Recommend $lpm_controls $oldsofts Safetyrisk C.Envirorisk_center#i.C_BioQ i.C_ID, vce(cluster C_ID) 
outreg2 using A7.xls, stats (coef tstat) bdec(3) tdec(2) addtext(Firm controls, Yes, Other Softs, Yes, Customer manager FE, Yes)  drop($lpm_controls  $oldsofts Safetyrisk i.C_ID) nocons
 
 

*	Table A8. Loan officers' biospheric-values–environmental-score interaction and their loan granting decision: standardized coefficients 
* Presented above after T7 Panel A



*	Table A9. Modeling the decisions of loan officers using biospheric-values interactions: logit specification 

* Continous environmental risk, centering
qui fvset base 4 L_BioQ
logit Outcome $lpm_controls Recommend $oldsofts Safetyrisk i.L_BioQ C.Envirorisk_center#i.L_BioQ, vce(cluster L_ID) 
outreg2 using A9.xls, stats (coef tstat) bdec(3) tdec(2) addtext(Firm controls, Yes, Other Softs, Yes)  drop($lpm_controls Recommend $oldsofts Safetyrisk) nocons

 

*	Table A10. When do loan officers exercise discretion? 

*	Panel A. Number of rejected loan applications by predicted approval decile and biospheric-values quintile 
*	Panel B. Number of approved loan applications by predicted approval decile and biospheric-values quintile
*	Panel C. Loan officer discretion as a function of customer managers' overoptimism
*	Panel D. Correlation between customer managers' overoptimism and loan performance measures


*****************************
* Fundamentals for analysis
*****************************

* Generate predicted approval probability
predict_grant_probs

* Divide predicted approval probability into 10 bins
xtile bin = p_approve_hat, nq(10) 

* Compute ingredients for the table, Panel A and B at the same time
tab2 bin Outcome if L_BioQ==0
tab2 bin Outcome if L_BioQ==1
tab2 bin Outcome if L_BioQ==2
tab2 bin Outcome if L_BioQ==3
tab2 bin Outcome if L_BioQ==4


************************************************************
* Customer manager overoptimism and loan officer discretion 
************************************************************

* List residual recommendations in the subsample of accepted loans belonging to bin 4
* A Simple count of the total number of obervations (4) and negative residual recommendations (4) tells the story
tab Resid_Rec if bin==4 & Outcome==1 

* List residual recommendations in the subsample of rejected loans belonging to at least bin 5
* A Simple count of the total number of obervations (33) and negative residual recommendations (33- 4 = 29) tells the story
tab Resid_Rec if bin>4 & Outcome==0 

* Test whether overoptimistic and pessimistic residual recommendations are equally likely
* Create first positive residual recommendation indicator
gen Resid_Rec_P = 0
replace Resid_Rec_P  = 1 if Resid_Rec > 0

* Perform a binomial test of whether overoptimistic and pessimistic residual recommendations are equally likely
bitest Resid_Rec_P == 0.5 if bin>4 & Outcome==0 

* Panel D. Calclulate correlation between residual recommendation and defaults and late payments or accepted loans
pwcorr Resid_Rec FinDefault FinLateRepayment if Outcome ==1, sig

**************************************************
* Two dimensions of loan officer discretion
**************************************************

* Joint distribution of residual recommendations and environmental risk for rejected loans belonging at least to predicted loan acceptance bin of 5 as a function of the biospheric values of the loan officer
* The point is to show that brown officers mostly use their discretion by rejecting green applitions, while green officers largely use their discretion to reject brown applications
tab2 Resid_Rec Envirorisk if bin>4 & Outcome ==0 & L_BioQ  == 0
tab2 Resid_Rec Envirorisk if bin>4 & Outcome ==0 & L_BioQ  == 1
tab2 Resid_Rec Envirorisk if bin>4 & Outcome ==0 & L_BioQ  == 2
tab2 Resid_Rec Envirorisk if bin>4 & Outcome ==0 & L_BioQ  == 3
tab2 Resid_Rec Envirorisk if bin>4 & Outcome ==0 & L_BioQ  == 4

* Test whether the environemntal risk of loan applicatnts for which brown and green officers apply discretion differs statistically significantly from one another
ttest Envirorisk if bin>4 & Outcome==0  & (L_BioQ==0 | L_BioQ==4), by(L_BioQ)

* Joint distribution of environmental scores and defaults or late payments of approved predicted approval bin 4 applications
tab2 Envirorisk FinDefault if Outcome ==1 & bin == 4
tab2 Envirorisk FinLateRepayment if Outcome ==1 & bin == 4



*	Table A11. Loan officers' biospheric-values–environmental-score interaction and their loan granting decisions: subsample analysis by loan approval probability terciles 

* Divide sample into terciles based on approval probablity
qui egen ApproveT = cut(p_approve_hat), group(3)

* Rerun loan pproval regression results separately for each tercile based on approval probablity

regress Outcome $lpm_controls Recommend $oldsofts Safetyrisk i.L_BioQ C.Envirorisk_center#i.L_BioQ if ApproveT==0, vce(cluster L_ID) 
outreg2 using A11.xls, stats (coef tstat) bdec(3) tdec(2) addtext(Firm controls, Yes, Other Softs, Yes)  drop($lpm_controls Recommend $oldsofts Safetyrisk) nocons

regress Outcome $lpm_controls Recommend $oldsofts Safetyrisk i.L_BioQ C.Envirorisk_center#i.L_BioQ if ApproveT==1, vce(cluster L_ID) 
outreg2 using A11.xls, stats (coef tstat) bdec(3) tdec(2) addtext(Firm controls, Yes, Other Softs, Yes)  drop($lpm_controls Recommend $oldsofts Safetyrisk) nocons

regress Outcome $lpm_controls Recommend $oldsofts Safetyrisk i.L_BioQ C.Envirorisk_center#i.L_BioQ if ApproveT==2, vce(cluster L_ID) 
outreg2 using A11.xls, stats (coef tstat) bdec(3) tdec(2) addtext(Firm controls, Yes, Other Softs, Yes)  drop($lpm_controls Recommend $oldsofts Safetyrisk) nocons


************************************************************************
* Important additional material related to A11 / cost of capital starts
* Effect on interest rate can be ignored
* Reported in text but not as table
************************************************************************

clear all
do "W:\Research\Current research\China\Analyses\data_initializing.do"

* Correlation between interest rate and environmental score, separately for brown and green officers
pwcorr Envirorisk ExecutiveAnnualInterestRate if Outcome ==1 & L_BioQ == 0, sig
pwcorr Envirorisk ExecutiveAnnualInterestRate if Outcome ==1 & L_BioQ == 4, sig

* Generate predicted approval probability
predict_grant_probs

* Correlation between predicted loan acceptance and interest rate
pwcorr p_approve_hat ExecutiveAnnualInterestRate if Outcome ==1, sig

******************************************************************
* Important additional material related to cost of capital ends
******************************************************************



*	Table A12. Loan outcomes for firms with male vs. female majority owners

clear all
do "W:\Research\Current research\China\Analyses\data_initializing.do"

* Descriptive statistics on female entrepreneurs
tab FemaleOwner

* Test for differences between male and female entrepreneurs
ttest Recommend, by(FemaleOwner)
ttest Outcome, by(FemaleOwner)
ttest ExecutiveAnnualInterestRate if Outcome == 1 , by(FemaleOwner)
ttest FinDefault if Outcome == 1 , by(FemaleOwner)


*	Table A13. Gender bias vs. green bias among loan officers 

clear all
do "W:\Research\Current research\China\Analyses\data_initializing.do"

qui fvset base 4 L_BioQ

regress Outcome $lpm_controls Recommend $oldsofts Safetyrisk C.Envirorisk_center L_Female FemaleOwner L_FemaleInteraction, vce(cluster L_ID) 
outreg2 using A13.xls, stats (coef tstat) bdec(3) tdec(2) addtext(Firm controls, Yes, Other Softs, Yes)  drop($lpm_controls Recommend $oldsofts Safetyrisk) nocons

regress Outcome $lpm_controls Recommend $oldsofts Safetyrisk i.L_BioQ C.Envirorisk_center#i.L_BioQ, vce(cluster L_ID) 
outreg2 using A13.xls, stats (coef tstat) bdec(3) tdec(2) addtext(Firm controls, Yes, Other Softs, Yes)  drop($lpm_controls Recommend $oldsofts Safetyrisk) nocons

regress Outcome $lpm_controls Recommend $oldsofts Safetyrisk i.L_BioQ C.Envirorisk_center#i.L_BioQ L_Female FemaleOwner L_FemaleInteraction, vce(cluster L_ID) 
outreg2 using A13.xls, stats (coef tstat) bdec(3) tdec(2) addtext(Firm controls, Yes, Other Softs, Yes)  drop($lpm_controls Recommend $oldsofts Safetyrisk) nocons


*	Table A14. Gender bias vs. green bias among customer managers 

clear all
do "W:\Research\Current research\China\Analyses\data_initializing.do"

qui fvset base 4 C_BioQ

regress Recommend $lpm_controls $oldsofts Safetyrisk C.Envirorisk_center C_Female FemaleOwner C_FemaleInteraction, vce(cluster C_ID) 
outreg2 using A14.xls, stats (coef tstat) bdec(3) tdec(2) addtext(Firm controls, Yes, Other Softs, Yes)  drop($lpm_controls  $oldsofts Safetyrisk) nocons

regress Recommend $lpm_controls $oldsofts Safetyrisk i.C_BioQ C.Envirorisk_center#i.C_BioQ, vce(cluster C_ID) 
outreg2 using A14.xls, stats (coef tstat) bdec(3) tdec(2) addtext(Firm controls, Yes, Other Softs, Yes)  drop($lpm_controls  $oldsofts Safetyrisk) nocons

regress Recommend $lpm_controls $oldsofts Safetyrisk i.C_BioQ C.Envirorisk_center#i.C_BioQ C_Female FemaleOwner C_FemaleInteraction, vce(cluster C_ID) 
outreg2 using A14.xls, stats (coef tstat) bdec(3) tdec(2) addtext(Firm controls, Yes, Other Softs, Yes)  drop($lpm_controls  $oldsofts Safetyrisk) nocons



*	Table A15. Additional descriptive statistics on customer managers and loan officers

*   Panel A. Customer managers: See output from Table 2 Panel A, which includes these additional descriptive statistics

*   Panel B. Loan officers: See Table 2 Panel B, which includes these additional descriptive statistics

*	Panel C. Correlations between customer managers' biospheric values and their other environmental traits (N = 190)
clear all
do "W:\Research\Current research\China\Analyses\data_initializing.do"
asdoc pwcorr C_Biosphericvalues C_Envvalues C_Envbeliefs C_Envinfo, sig append label title(Correlations between customer managers' biosphreric values and their other environmental traits)

*	Panel D. Correlations between loan officers' biospheric values and their other environmental traits (N = 57)
clear all
do "W:\Research\Current research\China\Analyses\data_initializing.do"
collapse (mean) L_Biosphericvalues L_Envvalues L_Envbeliefs L_Envinfo, by (L_ID)
asdoc pwcorr L_Biosphericvalues L_Envvalues L_Envbeliefs L_Envinfo, sig append label title(Correlations between loan officers' biosphreric values and their other environmental traits)



*	Table A16. Distinguishing between taste-based and statistical discrimination in loan officers' decision-making 

*	Panel A. Biospheric-values interactions
* Specification 1 comes from Table 7 Panel B Specification 1
* Specification 2: use residuals from beliefs alone
clear
do "W:\Research\Current research\China\Analyses\data_initializing.do"

reg L_Biosphericvalues L_Envbeliefs, vce(cluster L_ID)
qui predict Predicted_L_Bio
qui gen Resid_L_Bio = L_Biosphericvalues - Predicted_L_Bio
qui egen L_Resid_BioQ = cut(Resid_L_Bio), group(5)

qui fvset base 4 L_Resid_BioQ 
regress Outcome $lpm_controls Recommend $oldsofts Safetyrisk i.L_Resid_BioQ C.Envirorisk_center#i.L_Resid_BioQ, vce(cluster L_ID) 
outreg2 using A16a2.xls, stats (coef tstat) bdec(3) tdec(2) addtext(Firm controls, Yes, Other Softs, Yes)  drop($lpm_controls Recommend $oldsofts Safetyrisk) nocons

* Specification 3: use residuals from beliefs + environmental info
clear
do "W:\Research\Current research\China\Analyses\data_initializing.do"

reg L_Biosphericvalues L_Envbeliefs L_Envinfo, vce(cluster L_ID)
qui predict Predicted_L_Bio
qui gen Resid_L_Bio = L_Biosphericvalues - Predicted_L_Bio
qui egen L_Resid_BioQ = cut(Resid_L_Bio), group(5)

qui fvset base 4 L_Resid_BioQ 
regress Outcome $lpm_controls Recommend $oldsofts Safetyrisk i.L_Resid_BioQ C.Envirorisk_center#i.L_Resid_BioQ, vce(cluster L_ID) 
outreg2 using A16a3.xls, stats (coef tstat) bdec(3) tdec(2) addtext(Firm controls, Yes, Other Softs, Yes)  drop($lpm_controls Recommend $oldsofts Safetyrisk) nocons

*	Panel B. Green-beliefs interactions
* Specification 1: use beliefs as such 
clear
do "W:\Research\Current research\China\Analyses\data_initializing.do"
qui fvset base 4 L_EnvbeliefsQ

* Full sample, repeated from Table 8 Panel C Specification 1
regress Outcome $lpm_controls Recommend $oldsofts Safetyrisk i.L_EnvbeliefsQ C.Envirorisk_center#i.L_EnvbeliefsQ, vce(cluster L_ID) 
outreg2 using A16b1.xls, stats (coef tstat) bdec(3) tdec(2) addtext(Firm controls, Yes, Other Softs, Yes)  drop($lpm_controls Recommend $oldsofts Safetyrisk) nocons

* Specification 2: use residuals from biospheric values alone
clear
do "W:\Research\Current research\China\Analyses\data_initializing.do"
reg L_Envbeliefs L_Biosphericvalues, vce(cluster L_ID)
qui predict Predicted_L_Bel
qui gen Resid_L_Bel = L_Envbeliefs - Predicted_L_Bel
qui egen L_Resid_BelQ = cut(Resid_L_Bel), group(5)

qui fvset base 4 L_Resid_BelQ 
regress Outcome $lpm_controls Recommend $oldsofts Safetyrisk i.L_Resid_BelQ C.Envirorisk_center#i.L_Resid_BelQ, vce(cluster L_ID) 
outreg2 using A16b2.xls, stats (coef tstat) bdec(3) tdec(2) addtext(Firm controls, Yes, Other Softs, Yes)  drop($lpm_controls Recommend $oldsofts Safetyrisk) nocons

* Specification 3: use residuals from biospheric values + environmental info
clear
do "W:\Research\Current research\China\Analyses\data_initializing.do"
reg L_Envbeliefs L_Biosphericvalues L_Envinfo, vce(cluster L_ID)
qui predict Predicted_L_Bel
qui gen Resid_L_Bel = L_Envbeliefs - Predicted_L_Bel
qui egen L_Resid_BelQ = cut(Resid_L_Bel), group(5)

qui fvset base 4 L_Resid_BelQ 
regress Outcome $lpm_controls Recommend $oldsofts Safetyrisk i.L_Resid_BelQ C.Envirorisk_center#i.L_Resid_BelQ, vce(cluster L_ID) 
outreg2 using A16b3.xls, stats (coef tstat) bdec(3) tdec(2) addtext(Firm controls, Yes, Other Softs, Yes)  drop($lpm_controls Recommend $oldsofts Safetyrisk) nocons



*	Table A17. Loan granting decisions when loan officers hold below-median green beliefs 

clear
do "W:\Research\Current research\China\Analyses\data_initializing.do"
qui fvset base 4 L_BioQ

* Full sample, repeated from Table 7 Panel B Specification 1
regress Outcome $lpm_controls Recommend $oldsofts Safetyrisk i.L_BioQ C.Envirorisk_center#i.L_BioQ, vce(cluster L_ID) 
outreg2 using A17.xls, stats (coef tstat) bdec(3) tdec(2) addtext(Firm controls, Yes, Other Softs, Yes)  drop($lpm_controls Recommend $oldsofts Safetyrisk) nocons

* Below-median green beliefs subsample
regress Outcome $lpm_controls Recommend $oldsofts Safetyrisk i.L_BioQ C.Envirorisk_center#i.L_BioQ if L_Envbeliefs<3, vce(cluster L_ID) 
outreg2 using A17.xls, stats (coef tstat) bdec(3) tdec(2) addtext(Firm controls, Yes, Other Softs, Yes)  drop($lpm_controls Recommend $oldsofts Safetyrisk) nocons



*	Table A18. Placebo tests  using alternative banker traits

*	Panel A. Customer managers' overall recommendations by gender

* For creation of table
reg Recommend $lpm_controls $oldsofts Safetyrisk Envirorisk if C_Female==0 , vce(cluster C_ID) 
outreg2 using A18a.xls, stats (coef tstat) bdec(3) tdec(2) addtext(Industry FE, Yes, Firm type FE, Yes)  drop(i.Companyregtype i.Industrycode) nocons
reg Recommend $lpm_controls $oldsofts Safetyrisk Envirorisk if C_Female==1 , vce(cluster C_ID) 
outreg2 using A18a.xls, stats (coef tstat) bdec(3) tdec(2) addtext(Industry FE, Yes, Firm type FE, Yes)  drop(i.Companyregtype i.Industrycode) nocons			
				
* Analysis continues. Customer managers' recommendation regression, by gender, computing differences in key variables between genders

* Regressions for tests of differences in coefficients; drop clustering at the first stage

* The following fvset command is needed 
fvset base 1 Companyregtype

* Male group
reg Recommend $lpm_controls $oldsofts Safetyrisk Envirorisk if C_Female==0 
estimates store C_Female0

* Femaale group
reg Recommend $lpm_controls $oldsofts Safetyrisk Envirorisk if C_Female==1  
estimates store C_Female1
				
* Perform tests of differences for different variables
suest C_Female0 C_Female1, vce(cluster C_ID)

asdoc test [C_Female0_mean]Firmage = [C_Female1_mean]Firmage, append label title(Test of equality of customer managers' firm age coefficients, difference between men and women)
capture log using "W:\Research\Current research\China\Analyses\Output\myfile.smcl", text append
asdoc test [C_Female0_mean]logTotalAsset = [C_Female1_mean]logTotalAsset, append label title(Test of equality of customer managers' log total asset coefficients, difference between men and women)
capture log using "W:\Research\Current research\China\Analyses\Output\myfile.smcl", text append
asdoc test [C_Female0_mean]CurrentRatio_w = [C_Female1_mean]CurrentRatio_w, append label title(Test of equality of customer managers' crrent ratio coefficients, difference between men and women)
capture log using "W:\Research\Current research\China\Analyses\Output\myfile.smcl", text append
asdoc test [C_Female0_mean]DebtAssetRatio_w = [C_Female1_mean]DebtAssetRatio_w, append label title(Test of equality of customer managers' debt-to-asset ratio coefficients, difference between men and women)
capture log using "W:\Research\Current research\China\Analyses\Output\myfile.smcl", text append
asdoc test [C_Female0_mean]NetProfitRatio_w = [C_Female1_mean]NetProfitRatio_w, append label title(Test of equality of customer managers' net profit ratios coefficients, difference between men and women)
capture log using "W:\Research\Current research\China\Analyses\Output\myfile.smcl", text append
asdoc test [C_Female0_mean]SalesGrowth_w = [C_Female1_mean]SalesGrowth_w, append label title(Test of equality of customer managers' sales growth coefficients, difference between men and women)
capture log using "W:\Research\Current research\China\Analyses\Output\myfile.smcl", text append
asdoc test [C_Female0_mean]TruthfulnessandReasonableness = [C_Female1_mean]TruthfulnessandReasonableness, append label title(Test of equality of customer managers' truthfullness coefficients, difference between men and women)
capture log using "W:\Research\Current research\China\Analyses\Output\myfile.smcl", text append
asdoc test [C_Female0_mean]Repaymentability = [C_Female1_mean]Repaymentability, append label title(Test of equality of customer managers' repaymentability coefficients, difference between men and women)
capture log using "W:\Research\Current research\China\Analyses\Output\myfile.smcl", text append 
asdoc test [C_Female0_mean]RepaymentWillingness = [C_Female1_mean]RepaymentWillingness, append label title(Test of equality of customer managers' repayment willingness coefficients, difference between men and women)
capture log using "W:\Research\Current research\China\Analyses\Output\myfile.smcl", text append	 
asdoc test [C_Female0_mean]Safetyrisk = [C_Female1_mean]Safetyrisk, append label title(Test of equality of customer managers' safety risk coefficients, difference between men and women)
capture log using "W:\Research\Current research\China\Analyses\Output\myfile.smcl", text append
asdoc test [C_Female0_mean]Envirorisk = [C_Female1_mean]Envirorisk, append label title(Test of equality of customer managers' environmental risk coefficients, difference between men and women)
capture log using "W:\Research\Current research\China\Analyses\Output\myfile.smcl", text append
 
 
*	Panel B. Customer managers' overall recommendations by education

* For creation of table
reg Recommend $lpm_controls $oldsofts Safetyrisk Envirorisk if C_Highschool==0 , vce(cluster C_ID) 
outreg2 using A18b.xls, stats (coef tstat) bdec(3) tdec(2) addtext(Industry FE, Yes, Firm type FE, Yes)  drop(i.Companyregtype i.Industrycode) nocons
reg Recommend $lpm_controls $oldsofts Safetyrisk Envirorisk if C_Highschool==1 , vce(cluster C_ID) 
outreg2 using A18b.xls, stats (coef tstat) bdec(3) tdec(2) addtext(Industry FE, Yes, Firm type FE, Yes)  drop(i.Companyregtype i.Industrycode) nocons				

* Analysis continues. Customer managers' recommendation regression, by education, computing differences in key variables between education levels

* Regressions for tests of differences in coefficients; drop clustering at the first stage

* The following fvset command is needed 
fvset base 1 Companyregtype

* Highly educated group
reg Recommend $lpm_controls $oldsofts Safetyrisk Envirorisk if C_Highschool==0 
estimates store C_Highschool0

* High school group
reg Recommend $lpm_controls $oldsofts Safetyrisk Envirorisk if C_Highschool==1  
estimates store C_Highschool1
				
* Perform tests of differences for different variables
suest C_Highschool0 C_Highschool1, vce(cluster C_ID)

asdoc test [C_Highschool0_mean]Firmage = [C_Highschool1_mean]Firmage, append label title(Test of equality of customer managers' firm age coefficients, difference between high school and higher education)
capture log using "W:\Research\Current research\China\Analyses\Output\myfile.smcl", text append
asdoc test [C_Highschool0_mean]logTotalAsset = [C_Highschool1_mean]logTotalAsset, append label title(Test of equality of customer managers' log total asset coefficients, difference between high school and higher education)
capture log using "W:\Research\Current research\China\Analyses\Output\myfile.smcl", text append
asdoc test [C_Highschool0_mean]CurrentRatio_w = [C_Highschool1_mean]CurrentRatio_w, append label title(Test of equality of customer managers' crrent ratio coefficients, difference between high school and higher education)
capture log using "W:\Research\Current research\China\Analyses\Output\myfile.smcl", text append
asdoc test [C_Highschool0_mean]DebtAssetRatio_w = [C_Highschool1_mean]DebtAssetRatio_w, append label title(Test of equality of customer managers' debt-to-asset ratio coefficients, difference between high school and higher education)
capture log using "W:\Research\Current research\China\Analyses\Output\myfile.smcl", text append
asdoc test [C_Highschool0_mean]NetProfitRatio_w = [C_Highschool1_mean]NetProfitRatio_w, append label title(Test of equality of customer managers' net profit ratios coefficients, difference between high school and higher education)
capture log using "W:\Research\Current research\China\Analyses\Output\myfile.smcl", text append
asdoc test [C_Highschool0_mean]SalesGrowth_w = [C_Highschool1_mean]SalesGrowth_w, append label title(Test of equality of customer managers' sales growth coefficients, difference between high school and higher education)
capture log using "W:\Research\Current research\China\Analyses\Output\myfile.smcl", text append
asdoc test [C_Highschool0_mean]TruthfulnessandReasonableness = [C_Highschool1_mean]TruthfulnessandReasonableness, append label title(Test of equality of customer managers' truthfullness coefficients, difference between high school and higher education)	
capture log using "W:\Research\Current research\China\Analyses\Output\myfile.smcl", text append
asdoc test [C_Highschool0_mean]Repaymentability = [C_Highschool1_mean]Repaymentability, append label title(Test of equality of customer managers' repaymentability coefficients, difference between high school and higher education)
capture log using "W:\Research\Current research\China\Analyses\Output\myfile.smcl", text append
asdoc test [C_Highschool0_mean]RepaymentWillingness = [C_Highschool1_mean]RepaymentWillingness, append label title(Test of equality of customer managers' repayment willingness coefficients, difference between high school and higher education)
capture log using "W:\Research\Current research\China\Analyses\Output\myfile.smcl", text append
asdoc test [C_Highschool0_mean]Safetyrisk = [C_Highschool1_mean]Safetyrisk, append label title(Test of equality of customer managers' safety risk coefficients, difference between high school and higher education)
capture log using "W:\Research\Current research\China\Analyses\Output\myfile.smcl", text append
asdoc test [C_Highschool0_mean]Envirorisk = [C_Highschool1_mean]Envirorisk, append label title(Test of equality of customer managers' environmental risk coefficients, difference between high school and higher education)
capture log using "W:\Research\Current research\China\Analyses\Output\myfile.smcl", text append


*	Panel C. Customer managers' overall recommendations by age

* For creation of table
reg Recommend $lpm_controls $oldsofts Safetyrisk Envirorisk if C_Age < 30 , vce(cluster C_ID) 
outreg2 using A18c.xls, stats (coef tstat) bdec(3) tdec(2) addtext(Industry FE, Yes, Firm type FE, Yes)  drop(i.Companyregtype i.Industrycode) nocons
reg Recommend $lpm_controls $oldsofts Safetyrisk Envirorisk if C_Age > 29 & C_Age < 32  , vce(cluster C_ID) 
outreg2 using A18c.xls, stats (coef tstat) bdec(3) tdec(2) addtext(Industry FE, Yes, Firm type FE, Yes)  drop(i.Companyregtype i.Industrycode) nocons
reg Recommend $lpm_controls $oldsofts Safetyrisk Envirorisk if C_Age > 31 & C_Age < 35  , vce(cluster C_ID) 
outreg2 using A18c.xls, stats (coef tstat) bdec(3) tdec(2) addtext(Industry FE, Yes, Firm type FE, Yes)  drop(i.Companyregtype i.Industrycode) nocons
reg Recommend $lpm_controls $oldsofts Safetyrisk Envirorisk if C_Age > 34 & C_Age < 38 , vce(cluster C_ID) 
outreg2 using A18c.xls, stats (coef tstat) bdec(3) tdec(2) addtext(Industry FE, Yes, Firm type FE, Yes)  drop(i.Companyregtype i.Industrycode) nocons
reg Recommend $lpm_controls $oldsofts Safetyrisk Envirorisk if C_Age > 37 , vce(cluster C_ID) 
outreg2 using A18c.xls, stats (coef tstat) bdec(3) tdec(2) addtext(Industry FE, Yes, Firm type FE, Yes)  drop(i.Companyregtype i.Industrycode) nocons 				
				
* Analysis continues. Customer managers' recommendation regression, by age, computing differences in key variables between high and low age quintiles

* Regressions for tests of differences in coefficients; drop clustering at the first stage

* The following fvset command is needed 
fvset base 1 Companyregtype

* Lowest customer manager age quintile
reg Recommend $lpm_controls $oldsofts Safetyrisk Envirorisk if C_Age < 30 
estimates store C_AgeQ0

* Lowest customer manager age quintile
reg Recommend $lpm_controls $oldsofts Safetyrisk Envirorisk if C_Age > 37
estimates store C_AgeQ4
				
* Perform tests of differences for different variables
suest C_AgeQ0 C_AgeQ4, vce(cluster C_ID)

asdoc test [C_AgeQ0_mean]Firmage = [C_AgeQ4_mean]Firmage, append label title(Test of equality of customer managers' firm age coefficients, difference between top and bottom age quintiles)
capture log using "W:\Research\Current research\China\Analyses\Output\myfile.smcl", text append
asdoc test [C_AgeQ0_mean]logTotalAsset = [C_AgeQ4_mean]logTotalAsset, append label title(Test of equality of customer managers' log total asset coefficients, difference between top and bottom age quintiles)
capture log using "W:\Research\Current research\China\Analyses\Output\myfile.smcl", text append
asdoc test [C_AgeQ0_mean]CurrentRatio_w = [C_AgeQ4_mean]CurrentRatio_w, append label title(Test of equality of customer managers' crrent ratio coefficients, difference between top and bottom age quintiles)
capture log using "W:\Research\Current research\China\Analyses\Output\myfile.smcl", text append
asdoc test [C_AgeQ0_mean]DebtAssetRatio_w = [C_AgeQ4_mean]DebtAssetRatio_w, append label title(Test of equality of customer managers' debt-to-asset ratio coefficients, difference between top and bottom age quintiles)
capture log using "W:\Research\Current research\China\Analyses\Output\myfile.smcl", text append
asdoc test [C_AgeQ0_mean]NetProfitRatio_w = [C_AgeQ4_mean]NetProfitRatio_w, append label title(Test of equality of customer managers' net profit ratios coefficients, difference between top and bottom age quintiles)
capture log using "W:\Research\Current research\China\Analyses\Output\myfile.smcl", text append
asdoc test [C_AgeQ0_mean]SalesGrowth_w = [C_AgeQ4_mean]SalesGrowth_w, append label title(Test of equality of customer managers' sales growth coefficients, difference between top and bottom age quintiles)
capture log using "W:\Research\Current research\China\Analyses\Output\myfile.smcl", text append
asdoc test [C_AgeQ0_mean]TruthfulnessandReasonableness = [C_AgeQ4_mean]TruthfulnessandReasonableness, append label title(Test of equality of customer managers' truthfullness coefficients, difference between top and bottom age quintiles)
capture log using "W:\Research\Current research\China\Analyses\Output\myfile.smcl", text append
asdoc test [C_AgeQ0_mean]Repaymentability = [C_AgeQ4_mean]Repaymentability, append label title(Test of equality of customer managers' repaymentability coefficients, difference between top and bottom age quintiles)
capture log using "W:\Research\Current research\China\Analyses\Output\myfile.smcl", text append
asdoc test [C_AgeQ0_mean]RepaymentWillingness = [C_AgeQ4_mean]RepaymentWillingness, append label title(Test of equality of customer managers' repayment willingness coefficients, difference between top and bottom age quintiles)
capture log using "W:\Research\Current research\China\Analyses\Output\myfile.smcl", text append
asdoc test [C_AgeQ0_mean]Safetyrisk = [C_AgeQ4_mean]Safetyrisk, append label title(Test of equality of customer managers' safety risk coefficients, difference between top and bottom age quintiles)
capture log using "W:\Research\Current research\China\Analyses\Output\myfile.smcl", text append
asdoc test [C_AgeQ0_mean]Envirorisk = [C_AgeQ4_mean]Envirorisk, append label title(Test of equality of customer managers' environmental risk coefficients, difference between top and bottom age quintiles)
capture log using "W:\Research\Current research\China\Analyses\Output\myfile.smcl", text append


*	Panel D. Customer managers' overall recommendations by experience 

* For creation of table
reg Recommend $lpm_controls $oldsofts Safetyrisk Envirorisk if C_ExperienceinBank < 6 , vce(cluster C_ID) 
outreg2 using A18d.xls, stats (coef tstat) bdec(3) tdec(2) addtext(Industry FE, Yes, Firm type FE, Yes)  drop(i.Companyregtype i.Industrycode) nocons

reg Recommend $lpm_controls $oldsofts Safetyrisk Envirorisk if C_ExperienceinBank > 5 , vce(cluster C_ID) 
outreg2 using A18d.xls, stats (coef tstat) bdec(3) tdec(2) addtext(Industry FE, Yes, Firm type FE, Yes)  drop(i.Companyregtype i.Industrycode) nocons 				
				

* Analysis continues. Customer managers' recommendation regression, by experience, computing differences in key variables between high and low experience quintiles

* Regressions for tests of differences in coefficients; drop clustering at the first stage

* The following fvset command is needed 
fvset base 1 Companyregtype

* Lowest customer manager experience quintile
reg Recommend $lpm_controls $oldsofts Safetyrisk Envirorisk if C_ExperienceinBank < 6
estimates store C_ExperienceQ0

* Lowest customer manager experience quintile
reg Recommend $lpm_controls $oldsofts Safetyrisk Envirorisk if C_ExperienceinBank > 5
estimates store C_ExperienceQ4
				
* Perform tests of differences for different variables
suest C_ExperienceQ0 C_ExperienceQ4, vce(cluster C_ID)

asdoc test [C_ExperienceQ0_mean]Firmage = [C_ExperienceQ4_mean]Firmage, append label title(Test of equality of customer managers' firm age coefficients, difference between top and bottom experience quintiles)
capture log using "W:\Research\Current research\China\Analyses\Output\myfile.smcl", text append
asdoc test [C_ExperienceQ0_mean]logTotalAsset = [C_ExperienceQ4_mean]logTotalAsset, append label title(Test of equality of customer managers' log total asset coefficients, difference between top and bottom experience quintiles)
capture log using "W:\Research\Current research\China\Analyses\Output\myfile.smcl", text append
asdoc test [C_ExperienceQ0_mean]CurrentRatio_w = [C_ExperienceQ4_mean]CurrentRatio_w, append label title(Test of equality of customer managers' crrent ratio coefficients, difference between top and bottom experience quintiles)
capture log using "W:\Research\Current research\China\Analyses\Output\myfile.smcl", text append
asdoc test [C_ExperienceQ0_mean]DebtAssetRatio_w = [C_ExperienceQ4_mean]DebtAssetRatio_w, append label title(Test of equality of customer managers' debt-to-asset ratio coefficients, difference between top and bottom experience quintiles)
capture log using "W:\Research\Current research\China\Analyses\Output\myfile.smcl", text append
asdoc test [C_ExperienceQ0_mean]NetProfitRatio_w = [C_ExperienceQ4_mean]NetProfitRatio_w, append label title(Test of equality of customer managers' net profit ratios coefficients, difference between top and bottom experience quintiles)
capture log using "W:\Research\Current research\China\Analyses\Output\myfile.smcl", text append
asdoc test [C_ExperienceQ0_mean]SalesGrowth_w = [C_ExperienceQ4_mean]SalesGrowth_w, append label title(Test of equality of customer managers' sales growth coefficients, difference between top and bottom experience quintiles)
capture log using "W:\Research\Current research\China\Analyses\Output\myfile.smcl", text append
asdoc test [C_ExperienceQ0_mean]TruthfulnessandReasonableness = [C_ExperienceQ4_mean]TruthfulnessandReasonableness, append label title(Test of equality of customer managers' truthfullness coefficients, difference between top and bottom experience quintiles)
capture log using "W:\Research\Current research\China\Analyses\Output\myfile.smcl", text append	
asdoc test [C_ExperienceQ0_mean]Repaymentability = [C_ExperienceQ4_mean]Repaymentability, append label title(Test of equality of customer managers' repaymentability coefficients, difference between top and bottom experience quintiles)
capture log using "W:\Research\Current research\China\Analyses\Output\myfile.smcl", text append
asdoc test [C_ExperienceQ0_mean]RepaymentWillingness = [C_ExperienceQ4_mean]RepaymentWillingness, append label title(Test of equality of customer managers' repayment willingness coefficients, difference between top and bottom experience quintiles)
capture log using "W:\Research\Current research\China\Analyses\Output\myfile.smcl", text append
asdoc test [C_ExperienceQ0_mean]Safetyrisk = [C_ExperienceQ4_mean]Safetyrisk, append label title(Test of equality of customer managers' safety risk coefficients, difference between top and bottom experience quintiles)
capture log using "W:\Research\Current research\China\Analyses\Output\myfile.smcl", text append
asdoc test [C_ExperienceQ0_mean]Envirorisk = [C_ExperienceQ4_mean]Envirorisk, append label title(Test of equality of customer managers' environmental risk coefficients, difference between top and bottom experience quintiles)
capture log using "W:\Research\Current research\China\Analyses\Output\myfile.smcl", text append
 


*	Table A19. Loan officers' biospheric values and loan performance: robustness

* Basic regression, environmental score treated as continuous variable
regress FinDefault $lpm_controls Recommend $oldsofts Safetyrisk Envirorisk, vce(cluster L_ID) 
outreg2 using A19.xls, stats (coef tstat) bdec(3) tdec(2) addtext(Industry FE, Yes, Firm type FE, Yes) drop(i.Companyregtype i.Industrycode) nocons

* Add loan officer biospheric value quintiles, environmental score treated as continuous variable
regress FinDefault $lpm_controls Recommend $oldsofts Safetyrisk Envirorisk i.L_BioQ, vce(cluster L_ID) 
outreg2 using A19.xls, stats (coef tstat) bdec(3) tdec(2) addtext(Industry FE, Yes, Firm type FE, Yes) drop(i.Companyregtype i.Industrycode) nocons

* Environmental risk score interacted with loan officer biospheric value quintile
regress FinDefault $lpm_controls Recommend $oldsofts Safetyrisk i.L_BioQ C.Envirorisk_center#i.L_BioQ, vce(cluster L_ID) 
outreg2 using A19.xls, stats (coef tstat) bdec(3) tdec(2) addtext(Industry FE, Yes, Firm type FE, Yes) drop(i.Companyregtype i.Industrycode) nocons

* Environmental risk score interacted with loan officer biospheric value quintile, adding loan officer FEs
regress FinDefault $lpm_controls Recommend $oldsofts Safetyrisk C.Envirorisk_center#i.L_BioQ i.L_ID, vce(cluster L_ID) 
outreg2 using A19.xls, stats (coef tstat) bdec(3) tdec(2) addtext(Industry FE, Yes, Firm type FE, Yes, Loan officer FE, Yes) drop(i.Companyregtype i.Industrycode i.L_ID) nocons



*	Table A20. Loan officers' biospheric values and loan performance: loans overdue by less than 90 days

*	Panel A. Loan performance by loan officers' biospheric-values quintile
clear all
do "W:\Research\Current research\China\Analyses\data_initializing.do"
asdoc tab L_BioQ FinLateRepayment, append label title(Joint distribution of bad loans (late repayment) by loan officers' biospheric values) 


*	Panel B. Modeling loan performance with loan officers' biospheric values and interactions

qui fvset base 0 L_BioQ

* Basic regression, environmental score treated as continuous variable
regress FinLateRepayment $lpm_controls Recommend $oldsofts Safetyrisk Envirorisk Collateral Maturity2 ExecutiveAnnualInterestRate, vce(cluster L_ID) 
outreg2 using A20.xls, stats (coef tstat) bdec(3) tdec(2) addtext(Industry FE, Yes, Firm type FE, Yes) drop(i.Companyregtype i.Industrycode) nocons

* Add loan officer biospheric value quintiles, environmental score treated as continuous variable
regress FinLateRepayment $lpm_controls Recommend $oldsofts Safetyrisk Envirorisk Collateral Maturity2 ExecutiveAnnualInterestRate i.L_BioQ, vce(cluster L_ID) 
outreg2 using A20.xls, stats (coef tstat) bdec(3) tdec(2) addtext(Industry FE, Yes, Firm type FE, Yes) drop(i.Companyregtype i.Industrycode) nocons

* Environmental risk score interacted with loan officer biospheric value quintile
regress FinLateRepayment $lpm_controls Recommend $oldsofts Safetyrisk Collateral Maturity2 ExecutiveAnnualInterestRate i.L_BioQ C.Envirorisk_center#i.L_BioQ, vce(cluster L_ID) 
outreg2 using A20.xls, stats (coef tstat) bdec(3) tdec(2) addtext(Industry FE, Yes, Firm type FE, Yes) drop(i.Companyregtype i.Industrycode) nocons

* Environmental risk score interacted with loan officer biospheric value quintile, adding loan officer FEs
regress FinLateRepayment $lpm_controls Recommend $oldsofts Safetyrisk Collateral Maturity2 ExecutiveAnnualInterestRate C.Envirorisk_center#i.L_BioQ i.L_ID, vce(cluster L_ID) 
outreg2 using A20.xls, stats (coef tstat) bdec(3) tdec(2) addtext(Industry FE, Yes, Firm type FE, Yes, Loan officer FE, Yes) drop(i.Companyregtype i.Industrycode i.L_ID) nocons



*	Table A21. Subcomponents of biospheric values

*	Panel B. Modeling customer managers' biospheric-values–environmental-score interaction using different measures
* Continous environmental risk, centering
clear all
do "W:\Research\Current research\China\Analyses\data_initializing.do"

qui fvset base 4 C_BioResQ
regress Recommend $lpm_controls $oldsofts Safetyrisk i.C_BioResQ C.Envirorisk_center#i.C_BioResQ, vce(cluster C_ID) 
outreg2 using A21b.xls, stats (coef tstat) bdec(3) tdec(2) addtext(Firm controls, Yes, Other Softs, Yes)  drop($lpm_controls  $oldsofts Safetyrisk) nocons

qui fvset base 4 C_BioUniQ
regress Recommend $lpm_controls $oldsofts Safetyrisk i.C_BioUniQ C.Envirorisk_center#i.C_BioUniQ, vce(cluster C_ID) 
outreg2 using A21b.xls, stats (coef tstat) bdec(3) tdec(2) addtext(Firm controls, Yes, Other Softs, Yes)  drop($lpm_controls  $oldsofts Safetyrisk) nocons

qui fvset base 4 C_BioProQ
regress Recommend $lpm_controls $oldsofts Safetyrisk i.C_BioProQ C.Envirorisk_center#i.C_BioProQ, vce(cluster C_ID) 
outreg2 using A21b.xls, stats (coef tstat) bdec(3) tdec(2) addtext(Firm controls, Yes, Other Softs, Yes)  drop($lpm_controls  $oldsofts Safetyrisk) nocons

qui fvset base 4 C_BioPreQ
regress Recommend $lpm_controls $oldsofts Safetyrisk i.C_BioPreQ C.Envirorisk_center#i.C_BioPreQ, vce(cluster C_ID) 
outreg2 using A21b.xls, stats (coef tstat) bdec(3) tdec(2) addtext(Firm controls, Yes, Other Softs, Yes)  drop($lpm_controls  $oldsofts Safetyrisk) nocons

*	Panel A. Correlations between customer managers' mean biospheric values and its subcomponents
collapse (firstnm) C_BioQ C_BioResQ C_BioUniQ C_BioProQ C_BioPreQ, by (C_ID)
asdoc pwcorr C_BioQ C_BioResQ C_BioUniQ C_BioProQ C_BioPreQ, sig



*	Table A22. Loan granting decisions by industry greenness terciles 

clear all
do "W:\Research\Current research\China\Analyses\data_initializing.do"


*	Panel A: Coefficients and t-values 

* Perform regressions, first without loan officer fixed effects
qui fvset base 4 L_BioQ

* Brownest tercile of industries
reg Outcome $lpm_controls Recommend $oldsofts Safetyrisk i.L_BioQ C.Envirorisk_center#i.L_BioQ if Ind_env_score==1, vce(cluster L_ID) 
outreg2 using A22a.xls, stats (coef tstat) bdec(3) tdec(2) addtext(Firm controls, Yes, Other Softs, Yes)  drop($lpm_controls Recommend $oldsofts Safetyrisk) nocons

* Middle tercile of industries in terms of greenness
reg Outcome $lpm_controls Recommend $oldsofts Safetyrisk i.L_BioQ C.Envirorisk_center#i.L_BioQ if Ind_env_score==2, vce(cluster L_ID) 
outreg2 using A22a.xls, stats (coef tstat) bdec(3) tdec(2) addtext(Firm controls, Yes, Other Softs, Yes)  drop($lpm_controls Recommend $oldsofts Safetyrisk) nocons

* Greenest tercile of industries
reg Outcome $lpm_controls Recommend $oldsofts Safetyrisk i.L_BioQ C.Envirorisk_center#i.L_BioQ if Ind_env_score ==3, vce(cluster L_ID) 
outreg2 using A22a.xls, stats (coef tstat) bdec(3) tdec(2) addtext(Firm controls, Yes, Other Softs, Yes)  drop($lpm_controls Recommend $oldsofts Safetyrisk) nocons


*	Panel B: Standardized coefficients 

* Brownest tercile of industries
reg Outcome $lpm_controls Recommend $oldsofts Safetyrisk i.L_BioQ C.Envirorisk_center#i.L_BioQ if Ind_env_score==1, vce(r) beta
outreg2 using A22b.xls, stats (beta ) bdec(3) addtext(Firm controls, Yes, Other Softs, Yes)  drop($lpm_controls Recommend $oldsofts Safetyrisk) nocons

* Middle tercile of industries in terms of greenness
reg Outcome $lpm_controls Recommend $oldsofts Safetyrisk i.L_BioQ C.Envirorisk_center#i.L_BioQ if Ind_env_score==2, vce(r) beta 
outreg2 using A22b.xls, stats (beta ) bdec(3) addtext(Firm controls, Yes, Other Softs, Yes)  drop($lpm_controls Recommend $oldsofts Safetyrisk) nocons

* Greenest tercile of industries
reg Outcome $lpm_controls Recommend $oldsofts Safetyrisk i.L_BioQ C.Envirorisk_center#i.L_BioQ if Ind_env_score ==3, vce(r) beta
outreg2 using A22b.xls, stats (beta ) bdec(3) addtext(Firm controls, Yes, Other Softs, Yes)  drop($lpm_controls Recommend $oldsofts Safetyrisk) nocons



* Compute p-values for difference in key variable between high and low industry environmental risk terciles

* Regressions for tests of differences in coefficients; drop clustering at the first stage
* The following fvset command is needed 
fvset base 1 Companyregtype

* Brownest industries
reg Outcome Recommend Firmage logTotalAsset CurrentRatio_w DebtAssetRatio_w NetProfitRatio_w SalesGrowth_w i.Companyregtype ib1.Industrycode $oldsofts Safetyrisk i.L_BioQ C.Envirorisk_center#i.L_BioQ if Ind_env_score==1
estimates store Ind_env_score1

* Greenest industries
reg Outcome  Recommend Firmage logTotalAsset CurrentRatio_w DebtAssetRatio_w NetProfitRatio_w SalesGrowth_w i.Companyregtype ib1.Industrycode $oldsofts Safetyrisk i.L_BioQ C.Envirorisk_center#i.L_BioQ if Ind_env_score==3
estimates store Ind_env_score3

* Clustering comes into play here				
suest Ind_env_score1 Ind_env_score3, vce(cluster L_ID)

* Perform tests of differences for different variables
asdoc test [Ind_env_score1_mean]0.L_BioQ#c.Envirorisk_center = [Ind_env_score3_mean]0.L_BioQ#c.Envirorisk_center, append label title(Test of equality of loan officers' interaction coefficients, difference between greenest and brownest industries)
capture log using "W:\Research\Current research\China\Analyses\Output\myfile.smcl", text append



*	Table A23. Loan granting decisions by loan officer biospheric values quartiles 

clear all
do "W:\Research\Current research\China\Analyses\data_initializing.do"

qui egen L_BioT = cut(L_Biosphericvalues), group(4)

qui fvset base 3 L_BioT
regress Outcome $lpm_controls Recommend $oldsofts Safetyrisk i.L_BioT C.Envirorisk_center#i.L_BioT, vce(cluster L_ID) 
outreg2 using A23.xls, stats (coef tstat) bdec(3) tdec(2) addtext(Firm controls, Yes, Other Softs, Yes)  drop($lpm_controls Recommend $oldsofts Safetyrisk) nocons
regress Outcome $lpm_controls Recommend $oldsofts Safetyrisk C.Envirorisk_center#i.L_BioT i.L_ID, vce(cluster L_ID) 
outreg2 using A23.xls, stats (coef tstat) bdec(3) tdec(2) addtext(Firm controls, Yes, Other Softs, Yes, Loan officer FE, Yes)  drop($lpm_controls Recommend $oldsofts Safetyrisk i.L_ID) nocons



*	Table A24. Loan outcomes conditional on loan approval 

* Collateral
regress Collateral $lpm_controls Recommend  $oldsofts Envirorisk Safetyrisk  if (Outcome == 1), vce(cluster L_ID)
outreg2 using A24.xls, stats (coef tstat) bdec(5) tdec(2)  

* Interest rate 
regress ExecutiveAnnualInterestRate $lpm_controls Recommend  $oldsofts Envirorisk Safetyrisk  if (Outcome == 1), vce(cluster L_ID)
outreg2 using A24.xls, stats (coef tstat) bdec(5) tdec(2)  

* Maturity
regress Maturity2 $lpm_controls Recommend  $oldsofts Envirorisk Safetyrisk  if (Outcome == 1), vce(cluster L_ID)
outreg2 using A24.xls, append stats (coef tstat) bdec(3) tdec(2) 
 


*	Table A25. Treatment effects on environmental score, safety score, overall recommendation, and loan granting decision 

* LHS = Environmental score
regress Envirorisk $lpm_controls Green Brown, vce(cluster C_ID)
outreg2 using A25.xls, stats (coef tstat) bdec(3) tdec(2) addtext(Industry FE, Yes, Firm type FE, Yes) drop(i.Companyregtype i.Industrycode) nocons	

* LHS = Safety score
regress Safetyrisk $lpm_controls Green Brown, vce(cluster C_ID)
outreg2 using A25.xls, stats (coef tstat) bdec(3) tdec(2) addtext(Industry FE, Yes, Firm type FE, Yes) drop(i.Companyregtype i.Industrycode) nocons	

* LHS = Overall recommendation
regress Recommend $lpm_controls $oldsofts Safetyrisk Envirorisk Green Brown, vce(cluster C_ID) 
outreg2 using A25.xls, stats (coef tstat) bdec(3) tdec(2) addtext(Industry FE, Yes, Firm type FE, Yes) drop(i.Companyregtype i.Industrycode) nocons	

* LHS = Loan granted
regress Outcome $lpm_controls Recommend $oldsofts Safetyrisk Envirorisk Green Brown, vce(cluster L_ID) 
outreg2 using A25.xls, stats (coef tstat) bdec(3) tdec(2) addtext(Industry FE, Yes, Firm type FE, Yes) drop(i.Companyregtype i.Industrycode) nocons	


log close








